Grafting Poly(ethylene glycol) Epoxide to Amino-Derivatized Quartz

of pH (7.3−10.3) and reaction time (6−24 h) but was significantly influenced by reaction temperature (25−95 °C) and salt composition. PEG-g...
1 downloads 0 Views 219KB Size
Anal. Chem. 1996, 68, 3751-3757

Grafting Poly(ethylene glycol) Epoxide to Amino-Derivatized Quartz: Effect of Temperature and pH on Grafting Density Kazunori Emoto, J. Milton Harris, and James M. Van Alstine*

Department of Chemistry, University of Alabama in Huntsville, Huntsville, Alabama 35899

Silica-based materials such as glass and quartz are widely employed in a variety of analytical and separation methodologies. Examples include liquid chromatography,1-4 analytical microparticle electrophoresis,5,6 capillary electrophoresis,7-10 and highspeed DNA analysis and sequencing.7,11,12 The advantages and disadvantages of using such materials are well known.1,2 The

disadvantages primarily involve limited stability, variable performance, and strong solute adsorption (which reduces apparatus service life and causes phenomena such as peak tailing in chromatography and electrophoresis). Many authors have noted that quartz exhibits negatively charged siloxy (deprotonated silanol) groups over a wide pH range.1,2,4 Variation in quartz history, treatment, and hydration affect silanol group density and surface charge, thus reducing the effectiveness of various separation methods.1-10 Polymer coating of silica “masks” adsorption sites and stabilizes the underlying surface.2-6 In capillary electrophoresis, polymer coatings allow control over nonspecific adsorption and electroosmosis.5-7,13 Polymer-modified materials also find utilization in applications requiring control over wetting, nonspecific adsorption (e.g., fouling), and tethering of affinity ligands and other active groups.14-17 Hydrophilic, neutral polymers such as dextran,5,13,17 poly(ethylene glycol) (PEG),5,13,18-21 ethoxylated celluloses,21,22 and acrylamides23 are often used for such purposes. Attachment of polymers to colloid surfaces also reduces aggregation and improves their pharmaceutical utility.21,24 Given the above, there is a need to rapidly and nondestructively analyze polymer coated materials such as chromatography beads, biological particles, and capillaries. The ideal method should provide information about surface group pK’s and densities, polymer grafting density, and coating resistance to various environments, as well as effects related to variations in coupling method, chemistry, and polymer type. Analytical electrophoresis provides such information on polymer-coated particles, capillaries, and flat surfaces.5-7,16,25-30 The results correlate with various

(1) Xu, B.; Vermeullen, N. P. E. J. Chromatogr. 1988, 445, 1-28. (2) Anderson, D. J. Anal. Chem. 1995, 67, 475R-486R (in Clinical Chemistry Review 377R-524R). (3) Hanson, M.; Kurganov, A.; Unger, K. K.; Davankov, V. A. J. Chromatogr. 1993, 656, 369-380. (4) Schomburg, G. Trends. Anal. Chem. 1991, 10, 163-169. (5) Van Alstine, J. M.; Burns, N. L.; Riggs, J. A.; Holmberg, K.; Harris, J. M. Colloids Surf. A 1993, 77, 149-158. (6) Nordt, F. J.; Knox, R. J.; Seaman, G. V. F. In Hydrogels for Medical and Related Applications; Andrade, J. D., Ed.; ACS Symposium Series 31; American Chemical Society: Washington, DC, 1976; Chapter 17, pp 225-240. (7) Grossman, P. D. In Capillary Electrophoresis: Theory and Practice; Grossman, P. D.; Colburn, J. C., Eds.; Academic Press: New York, 1992; pp 3-43. (8) Zhao, Z.; Malik, A.; Lee, M. L. Anal. Chem. 1993, 65, 2747-2752. (9) O’Neill, K.; Shao, X.; Zhao, Z.; Malik, A.; Lee, M. L. Anal. Biochem. 1994, 222, 185-189. (10) Gilges, M.; Kleemiss, M. H.; Schomburg, G. Anal. Chem. 1994, 66, 20382046. (11) Xu, Y. Anal. Chem. 1995, 67, 463R-474R (in Clinical Chemistry Review, 377R-524R). (12) Fung, E. N.; Yeung, E. S. Anal. Chem. 1995, 67, 1913-1919.

(13) Herren, B. J.; Shafer, S. G.; Van Alstine, J. M.; Harris, J. M.; Snyder, R. S. J. Colloid Interface Sci. 1987, 115, 46-55. (14) Yoshinaga, K.; Shafer, S. G.; Harris, J. M. J. Bioact. Compat. Polym. 1987, 2, 49-56. (15) Mrksich, M.; Whitesides, G. M. Trends Biotechnol. 1995, 13, 228-235. (16) Scouten, W. H.; Luong, J. H.; Brown, R. S. Trends Biotechnol. 1995, 13, 178-185. (17) Boyce, J. F.; Hovanes, B A.; Harris, J. M.; Van Alstine, J. M.; Brooks, D. E. J. Colloid Interface Sci. 1992, 149, 153-161. (18) Burns, N. L.; Van Alstine, J. M.; Harris, J. M. Langmuir 1995, 11, 27682776. (19) Jeon, S. I.; Lee, J. H.; Andrade, J. D.; de Gennes, P. J. J. Colloid Interface Sci. 1990, 142, 149-158. (20) Lin, Y. S.; Hlady, V.; Go ¨lander, C. G. Colloids Surf. B 1994, 3, 49-62. (21) Harris, J. M., Ed. Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications; Plenum: New York, 1992. (22) Malmsten, M.; Lindman, B.; Holmberg, K.; Brink, C. Langmuir 1991, 7, 2412-2413. (23) Hjerte´n, S. J. Chromatogr. 1985, 347, 191-198. (24) Delgado, C.; Francis, G. E.; Fisher, D. Crit. Rev. Ther. Drug Carrier Syst. 1992, 9, 249-304.

Microparticle capillary electrophoresis was used to characterize the surface of quartz capillaries grafted with the glycidyl ether of poly(ethylene glycol) (E-PEG). Site dissociation modeling of capillary electrokinetic behavior provided estimates of surface group pK and density, plus the distance (d) from the surface to the hydrodynamic plane of shear. Native quartz appeared to possess silanol groups of pK 3.6 and 6.9 whose surface densities varied with quartz treatment. Aminopropylsilane derivatization of quartz silanol groups in toluene yielded a coating which was stable (>6 h) at pH 10.3 and 60 °C. Aqueous grafting of E-PEG to this surface was relatively independent of pH (7.3-10.3) and reaction time (6-24 h) but was significantly influenced by reaction temperature (25-95 °C) and salt composition. PEG-grafted capillaries exhibited greatly reduced electroosmosis from pH 2 to 11. Significant grafting could be obtained under mild conditions (6 h, 35 °C, 0.4 M K2SO4, pH 6.9). These results suggest that PEG chains increasingly extend normal to a surface as their grafting density increases, and that PEG conformation influences grafting density. The methods described should aid the use of PEG-coated surfaces in a variety of applications.

S0003-2700(96)00114-X CCC: $12.00

© 1996 American Chemical Society

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996 3751

Figure 1. Schematic treatment of quartz surface with aminopropylsilane and E-PEG.

aspects of coupling chemistry,5,6,13,18,29-32 streaming potentials,33,34 and the ability of polymer coatings to alter partition coefficients, reduce nonspecific protein adsorption, and alter surface wetting.5,13,17,35,36 Glycidyl ether (epoxide)-functionalized PEGs (E-PEGs) have been used to coat surfaces and tether enzymes to glass particles.35-38 The present work involves use of microparticle capillary electrophoresis, paired with electrokinetic modeling18,27,33 to examine the utility of E-PEG for covalently coating aminosilane-modified quartz (Figure 1). Electrophoresis provided information related to quartz surface chemistry and polymer grafting density. Altering reaction pH, time, temperature, and ionic strength allowed the identification of mild conditions which yield surfaces of varying PEG grafting density. EXPERIMENTAL SECTION Chemicals were ACS grade or better. Solutions were prepared immediately prior to use. High-purity, ion-exchange-treated (25) Fornasiero, D.; Fengsheng, L.; Ralston, J. J. Colloid Interface Sci. 1994, 164, 345-354. (26) Hunter, R. J. Zeta Potential in Colloid SciencesPrinciples and Applications; Academic Press: New York, 1981. (27) Healy, T. W.; White, L. R. Adv. Colloid Interface Sci. 1978, 9, 303-345. (28) Uchida, E.; Uyama Y.; Ikada, Y. Langmuir 1994, 10, 1193-1198. (29) Koopal, L. K.; Hlady, V.; Lyklema, J. J. Colloid Interface Sci. 1987, 121, 49-61. (30) Eremenko, V. V.; Malysheva, M. L.; Rusina, O. D.; Kutsevol, N. V.; Zeltonozskaya, T. B. Colloids Surf. A 1995, 98, 19-24. (31) Norde, W.; Lyklema, J. In Surface and Interfacial Aspects of Biomedical Polymers. Vol. 2, Protein Adsorption; Andrade, J. D., Ed.; Plenum Press: New York, 1985; pp 241-262. (32) Lun, B.; Xie, J.; Lu, C.; Wu, C.; Wei, Y. Anal. Chem. 1995, 67, 83-87. (33) Scales, P. J.; Grieser, F.; Healy T. W.; White, L. R.; Chan, D. Y. C. Langmuir 1992, 8, 965-974. (34) Zembala, M.; De´jardin, P. Colloids Surf. B 1994, 3, 119-129. (35) O ¨ sterberg, E.; Bergstro ¨m, K.; Holmberg, K.; Riggs, J. A.; Van Alstine, J. M.; Shuman, T. P.; Burns, N. L.; Harris, J. M. Colloids Surf. A 1993, 77, 159-169. (36) O ¨ sterberg, E.; Bergstro ¨m, K.; Holmberg, K.; Schuman, T. P.; Riggs, J.; Burns, N. L.; Van Alstine, J. M.; Harris, J. M. J. Biomed. Mater. Res. 1995, 29, 741-747. (37) Bergstro ¨m, K.; O ¨ sterberg, E.; Holmberg, K.; Hoffman, A. S.; Schuman, T. P.; Kozlowski, A.; Harris, J. M. J. Biomater. Sci. Polym. Ed. 1994, 6, 123132. (38) Bergstro ¨m, K.; Holmberg, K. Biotechnol. Bioeng. 1991, 38, 952-955.

3752

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

(Barnsted Ultrapure System) water was used throughout. Solution conductivites were controlled via use of a digital conductivity meter (Radiometer Copenhagen). Reaction and stock solution pH was monitored at regular intervals with a digital pH meter (Dow Corning). Quartz Glass Capillaries. Quartz capillaries of 2.0 mm i.d., 4.0 mm o.d., and 100 mm length (Vitro Dynamics Inc.) were used as they minimize extraneous fluid flow and provide the large lengthto-diameter ratio needed for effective electroosmosis measurement.5,6,13 Unless noted otherwise, capillaries were used fresh or stored in distilled water at room temperature and rinsed copiously with water before use. Quartz capillaries were cleaned prior to coating by sequential exposure to 1% (w/w) NaOH (60 °C, 10 min), 3% (w/w) HCl (60 °C, 10 min), and boiling 30% (v/v) H2O2 for 1 h, before being rinsed repeatedly with distilled water. Aminopropylsilane (APS) Coating. Capillaries were dried (22 °C, 4 h, 10-2 Torr) and then immersed for 4 h in 2% or 5% (v/v) (3-aminopropyl)triethoxysilane (APS) in anhydrous toluene (Aldrich). The capillaries were then rinsed with toluene and cured (190 °C, 12 h, 10-2 Torr) to create a stable silane monolayer. Control capillaries were subjected to the above conditions in the absence of APS. Some APS-modified capillaries were subjected to 6 h of hydrothermal treatment at various temperatures (25 °C95 °C) in 0.05 M sodium phosphate, pH 7.3, or in 0.05 M sodium carbonate at pH 10.3. PEG Grafting and Control Adsorption. Control capillaries were prepared by subjecting clean quartz capillaries to APS curing conditions (see above) and then adsorbing PEG of 25 000 Da (Shearwater Polymers) via 6 h, 60 °C exposure to a 10% (w/v) polymer solution in 0.05 M sodium phosphate, pH 7.3, followed by rinsing with water. The difunctional glycidyl ether (epoxide) of 3400 Da PEG from Shearwater Polymers (E-PEG) was grafted to APS-modified capillaries under a variety of conditions: i.e., 6-24 h at 25-95 °C as a 10% (w/v) solution in the above 0.05 M buffers (pH 7.310.3) or 6 h at 35 °C in 0.45 M K2SO4, pH 6.9. Electrophoresis Measurements. Microparticle analytical electrophoresis was used to determine the variation in electroosmosis associated with native and modified capillaries from pH 2 to 11 in 7.5 mM medium or from pH 3 to 11 in 1 mM medium. At pH 3, 5.8, and 11, the 7.5 mM solutions had conductivities of 1.170, 0.890, and 0.990 µS‚cm-1, respecively, while the 1.0 mM solutions had conductivities of 0.420, 0.320, and 0.425 µS‚cm-1, respectively. The mean electrophoretic mobility of detergent-free, 1 µm diameter, sulfated polystyrene latex (PSL) microspheres (charge density 0.003 22 mEquiv‚g-1, Interfacial Dynamics Corp.) was measured in a modified Rank analytical microelectrophoresis system (Rank Bros., Cambridge, UK) with platinum electrodes (6,13) and digital multimeters (J. Fluke Co.). Care was taken to reduce exposure of particles to extremes of pH.35 A 2 mm diameter quartz capillary was mounted horizontally between two poly(methyl methacrylate) blocks containing chambers for electrodes, as shown in Figure 2a. The electrophoresis chamber was filled with a suspension of PSL beads in 7.5 or 1.0 mM NaCl medium, whose pH had been adjusted by adding either 7.5 or 1.0 mM NaOH or HCl, and electrodes were tightly set into the blocks. The chamber was immersed in a water bath (25.0 ( 0.1 °C). A 400× water immersion microscope equipped with three-axis distance micrometers and with an ocular graticule was

theory can be used to relate the electroosmotic contribution to capillary wall potential ζ, surface charge, and pH.18 The electrophoretic and electroosmotic contributions of particle mobility in the capillary26 are given by

U(r) ) (Up - Uo) + 2(r/ro)2 Uo

(1)

where U(r) is the observed mobility of the particle at a distance r from the center of a capillary with radius ro, and Up and Uo are electrophoretic and electroosmotic mobility, respectively. Observed mobility versus r (Figure 2c) indicates electroosmotic mobility, which can be modeled and fit as a parabola. Based on Gouy-Chapman theory, the electrostatic potential (Ψ) a distance x from the charged surface in a univalent electrolyte solution is given by

tanh

( ) ( ) (x ) 2e2n0 x kBT

eΨ0 eΨ ) tanh exp 4kBT 4kBT

(2)

where e is the Coulombic charge of an electron, kB the Boltzmann constant, T the absolute temperature,  the permittivity of the medium, n0 the solution ion concentration, and Ψ0 the surface potential (i.e., x ) 0). If F represents the charge density of the medium, then Ψ0 is related to total surface charge (σ0) by



σ0 ) Figure 2. (a) Rank microparticle electrophoresis apparatus modified for capillaries, (b) electroosmotic and compensatory flows in capillary, and (c) parabola of particle electrophoretic mobility versus position across capillary diameter.

used to visually determine the mobility of PSL particles subject to a 40 V (4 V‚cm-1) dc potential. Particle mobility was measured at 8-16 locations across the capillary diameter (Figure 2b,c). Particle mobility at each capillary location was obtained by averaging eight measurements, with the field reversed between measurements in order to avoid electrode polarization.6,13 The observed particle mobility consists of electrophoretic and electroosmotic contributions. The latter, which is not affected by slight variations in sulfated particle mobility at low pH,5,13,35 is due to surface charge groups located at the capillary wall. In the sealed apparatus (Figure 2a), wall electroosmosis is compensated by a return flow down the center of the capillary (Figure 2b), such that a plot of particle mobility versus position across the capillary diameter gives a parabolic distribution (Figure 2c). Only at the “stationary level”, where electroosmotic and compensatory flows cancel (Figure 2b), is particle mobility independent of electroosmosis. Electroosmosis, which provides information on capillary surface groups, can be taken as the second-order term of a secondorder curve fit of particle mobility versus position across the capillary (Figure 2c).6,18,26 Data modeling and theory are detailed in the next two sections. Electrokinetic Theory and Equations. Various materials, including quartz, exhibit ionized surface groups when in contact with an aqueous solution. The resulting surface charge induces an electrostatic potential related to an unbalanced ionic charge distribution in the solution (electric double layer). The electrophoretic and electroosmotic contributions of particle mobility under the above conditions are well known.26 Gouy-Chapman



0

( )

F dx ) -x8n0kBT sinh

eΨ0 2kBT

(3)

Healy and White27,33 described the charge of a uniformly charged planar surface as a function of discrete surface ionogenic sites and their dissociation constants (pK’s)

∑ [A

σ0 ) - e

i]

-

i

∑ i

∑ [BH

+e

[

j]

+

j

-eNai

)

] [ +

1 + 10pKai-pHs

∑ j

eNbj

]

1 + 10pHs-pKbj

(4)

where Na and Nb are the surface densities of i distinct deprotonated acidic (A-) and j distinct protonated basic (BH+) groups, with dissociation constants Ka and Kb, respectively. The term pHs represents the concentration of hydrogen ion at the charged surface. Solving the Smoluchowski equation gives the relation between electroosmosis and ζ potential at the hydrodynamic plane of shear, Ψ(d),

U0 )

vz0 ζ  ) - Ψ(d) ) Ez η η

(5)

where vz0 is the electroosmotic velocity, η is coefficient of viscosity and Ez is the applied electric field. Electrokinetic Modeling and Figures. The pK and density of functional groups were deduced by matching calculated and observed electroosmosis profiles, based on eqs 2-5 above and the Boltzman equation.18 A best fit of calculated and observed electroosmosis was achieved for each capillary over a wide range of pH. The solid lines in Figures 3, 5, 6, and 7 are modeling profiles for experimental results, which are indicated by symbols. Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

3753

Figure 4. (a) Schematic drawing of possible isolated, associated, and PEG-associated silanol groups on quartz surface. (b) Possible mechanism for apparent increase in negatively charged groups following hydrothermal treatment of APS-coated quartz. Table 1. Site-Dissociation Modeling of Native and APS-Modified Quartz Figure 3. Electroosmosis profiles at 7.5 mM of (O) native, (4) APSmodified, and (2) APS-modified and then hydrothermally treated (6 h, 95 °C, 0.05 M sodium phosphate, pH 7.3) capillaries. Solid lines are modeling fits to experimental data shown by symbols.

Error bars were approximately the same size as the symbols. Each plotted symbol represents a parabolic distribution associated with one pH and ionic strength, deduced from over 100 measurements conducted at various places in each capillary (Figure 2c). Values for electroosmotic mobility were deduced from a least-squares curve fit of data points using Cricket Graph software and a MacIntosh computer. Results were reproducible, even when experiments were conducted several months apart.5,18 Experimental electrokinetic data were “site dissociation” modeled27 over a broad range of pH (i.e., surface charge and ζ), on the basis of eqs 2-4.18 “Effective” values for Na, Nb, pK’s, and d were used in combination with Theorist software to calculate pHdependent electroosmosis profiles which fit experimental data. Many effective values could be assigned from the literature.18 A priori modeling experiments, where parameters were determined via least-squares fit with Matlab software, to p < 0.01 confidence levels, match the results reported here (not shown). Tables 1 and 2 summarize effective pK’s and surface densities of ionogenic groups, plus d values for variously treated quartz capillaries (discussed below). RESULTS AND DISCUSSION Native Quartz Surfaces. Figure 3 provides a plot of electroosmosis versus pH in a 2 mm quartz capillary, before and after coating with aminopropylsilane (Figure 1). The increase in electroosmosis with pH for native quartz matches that observed previously for both 2 mm diameter capillaries5,6 and the smaller diameter capillaries used in capillary electrophoresis.7 Many authors have noted that quartz exhibits negatively charged siloxy (deprotonated silanol) groups over a wide pH range.1,2,6 Estimates of surface silanol density typically approximate 5 groups‚nm-2, with siloxy pK estimates ranging from 5.8 to 7.2.1,18 This variation in pK is consistent with two types of surface silanols (Figure 4a).18 “Isolated” silanol groups are expected to be relatively acidic and may correlate with the “strongly reactive” silanols known to influence solute adsorption in liquid chromatography.2 “Associated” silanols (Figure 4a) exhibit higher pK’s as hydrogen bonding hinders deprotonation. Modeling of electrokinetic results (Table 3754 Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

surface native Quartz APS (0% v/v)a APS (2% v/v) APS (5% v/v) APS (5% v/v), hydrothermal 1b APS (5% v/v), hydrothermal 2b

surface density of charged groups per nm2 SiOHi SiOHa NH2 (pK 3.6)c (pK 6.9)c (pK 10.0)

shear plane distance,d d (nm)

0.03 0.10 0.15 0.07 0.18

5.00 0.50 0.18 0.20 0.23

0.00 0.00 0.20 0.20 0.20

0.9 0.9 1.3 1.3 1.3

0.18

0.23

0.20

1.3

a Control. b Hydrothermal treatment 1 is 6 h, 95 °C, 0.05 M sodium phosphate, pH 7.3. Treatment 2 is 6 h, 60 °C, 0.05 M sodium carbonate, pH 10.3. c SiOHi represents isolated silanol group, and SiOHa represents associated silanol group. d Distance from surface to electrohydrodynamic plane of shear.

1) is consistent with two charged groups of pK 3.6 and 6.9 and surface densities of 0.03 and 5.00 (charged groups‚nm-2). These values match those reported earlier for variously cleaned quartz surfaces. The d value of 0.9 nm (Table 1) may reflect surface roughness and “bound water”.18 Variation in quartz history, treatment, and hydration is expected to alter silanol group density, the ratio of isolated and associated groups, surface charge, and electroosmosis versus pH profiles (Table 1). This, in turn, is expected to affect surface wetting, solute adsorption, and other phenomena of practical significance.1-4 Amino-Modified Quartz Surfaces. Grafting aminopropylsilane onto native quartz (Figure 1) results in negative electroosmosis (net positive surface charge) at low pH where silanol groups are protonated (Figure 3). At the point of zero electroosmosis (PZE), the negative charge contribution from deprotonated silanol groups is balanced by the positive contribution of amino groups. PZE varies with the pK and density of charged surface groups. At higher pH, negative siloxy groups predominate and positive electroosmosis is observed. APS derivatization of quartz (Figure 1) is self-catalyzed by the amino groups present. This allows the reaction to occur in anhydrous toluene at surfaces containing residual bound water.39,40 Modeling results suggest that charged amino groups are present (39) Vandenberg, E. T.; Bertilsson, L.; Liedberg, B.; Uvdal, K.; Erlandsson, R.; Elwing, H.; Lundstro ¨m, I. J. Colloid Interface Sci. 1991, 147, 103-118. (40) Caravajal, S. G.; Lyden, D. E.; Quinting, G. R.; Maciel, G. E. Anal. Chem. 1988, 60, 1776-1786.

Figure 5. Electroosmosis profiles at 7.5 mM for APS-coated capillaries grafted with E-PEG in sodium carbonate solution at pH 7.3 for 6 h at (b) 25 °C and (O) 60 °C. Solid lines are modeling fits to experimental data shown by symbols.

on the surface at 0.2 groups‚nm-2 and that d is increased to 1.3 nm (Table 1). Increasing APS reaction concentration from 2% to 5% (v/v) did not alter these values. Apparently, excess APS was removed by the toluene rinse prior to curing. The density of associated silanol groups was reduced beyond that expected from the APS reaction conditions. Control experiments conducted without APS indicate that this is due to surface dehydration (silyl ether bond formation) during curing (Table 1). The stability of the surface-grafted APS groups was tested via 6 h “hydrothermal” exposure to 0.05 M, pH 7.3, salt solution at 95 °C, or pH 10.3 solution at 60 °C. In both cases, amino group surface density was apparently unaffected (Table 1, Figure 3), as evidenced by constant electroosmosis at pH 2.0 (where surface silanol groups are protonated and do not contribute to electroosmosis). Enhanced electroosmosis from pH 3 to 10 and a shift in PZE from pH 7.2 to 5.8 (Table 1, Figure 3) suggest that the treatment increased silanol surface density. These results are not related to surface oxidation or reaction of the APS-modified surface with carbon dioxide, as the capillaries used were stored at 4 °C under nitrogen. The apparent increase in siloxy groups may be due to hydrolysis of silyl ether linkages or removal of APS molecules bound electrostatically (via their amino groups) to surface siloxy groups (Figure 4b).32,39,40 Adsorption and Grafting of PEG to Capillaries. APS-modified capillaries were covlently grafted with PEG by exposure to 10% (w/w) solutions of E-PEG for 6 h in 0.05 M sodium phosphate, pH 7.3 (Figure 1). Under the conditions studied (see Experimental Section), the difunctional E-PEGs are expected to react via one or both end groups, with few unreacted epoxide groups oxidizing to form ionogenic groups (manuscript in preparation). The ability of PEG and related polymer coatings to mask surface charge expression and reduce capillary electroosmosis or particle electrophoretic mobility is well known (see introduction). In the present study, E-PEG-coated capillaries exhibited significant reductions in electroosmosis (Figures 5-7). The reduction in electroosmosis varied with reaction temperature over the range studied (25-95 °C). Coatings applied at 25 °C and 35 °C yielded comparable reductions in electroosmosis measured in 7.5 mM salt solution (Table 2). Coatings applied at 60 °C (Figure

Figure 6. Electroosmosis profiles at 1.0 mM of APS-modified capillaries grafted with E-PEG in sodium carbonate solution at pH 7.3 for 6 h at (O) 60 °C and (9) 95 °C. Solid lines refer to modeling fits.

Figure 7. Electroosmosis profiles at 7.5 mM for APS-modified capillaries grafted with E-PEG at 60 °C for 6 h in 0.05 M sodium phosphate pH 7.3 (O) or sodium carbonate at pH 8.3 (0), 9.3 (b), or 10.3 (9). Solid line refers to modeling fit to pH 7.3 data.

5) and 95 °C (not shown) were also comparable and much more effective than the coatings applied at lower temperatures. Differences in the 60 and 95 °C coatings were further investigated using 1 mM solutions, whose lower ionic strength yielded more pronounced electroosmosis and improved ability to evaluate d values greater than 5 nm.18,26,29,31 The 95 °C coating was found to be more effective than the 60 °C coating at reducing electroosmosis in 1 mM solution (Figure 6). Modeling results were similar for both the 1 and 7.5 mM conditions (Table 2) and are shown as solid lines in the figures. Experimental results (represented as plotted symbols) are best modeled with amino group density and pK kept constant, silanol density and d increasing with reaction temperature, and silanol group pK increasing as PEG becomes localized at the surface (Table 2). The constancy of amino group density and pK is in keeping with the apparent stability of the grafted APS. The apparent increase in silanol density with reaction temperature may be due to surface hydrolysis, as Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

3755

Table 2. Site-Dissociation Modeling of Quartz Coated with Adsorbed or Grafted PEG

reaction PEG 25000a E-PEG 3400b 25 °C 35 °C 60 °C 95 °C E-PEG 3400c 35 °C

surface density of charged groups per nm2 SiOHa NH2 SiOHi (pK 4.1)d (pK 8.0)d (pK 10.0)

shear plane distancee d (nm)

0.10

0.50

0.00

1.1

0.17 0.17 0.18 0.18

0.20 0.20 0.23 0.23

0.20 0.20 0.20 0.20

2.4 2.4 7.0 11.0

0.18

0.23

0.20

7.0

a Control adsorbed 0% (v/v) APS-exposed quartz (Table 1), for 6 h in 0.050 M sodium phosphate, pH 7.3, at 60 °C. b For 5% (v/v) APScoated quartz (Table 1), 6 h, 0.050 M sodium phosphate, pH 7.3, at temperature indicated. c Reaction 6 h, 0.45 M K2SO4, pH 6.9, at 35 °C on 5% (v/v) APS-coated quartz (Table 1). d SiOHi represents isolated silanol group, and SiOHa represents associated silanol group. E-PEG 95 °C data modeled with silanol pK’s of 3.6 and 6.9; all other data modeled with silanol pK’s of 4.1 and 8.0. e Distance from surface to electrohydrodynamic plane of shear. Modeling results similar for 7.5 and 1.0 mM solutions.

discussed above. The increases in both d and silanol group pK values are discussed below. The ability of terminally grafted PEG coatings to reduce electroosmosis can vary with PEG molecular mass.5,13 At low ionic strength, d values are expected to describe a region where fluid movement is restricted, and that therefore closely approximate the hydrodynamic thickness of the polymer coating.29 The authors previously noted that coatings of approximately equal grafting density, produced with PEGs of 500-35 000 Da, exhibited d values from 1.6 to >10 nm, respectively.18 The experimental data (Figures 5 and 6) for 3400 Da E-PEG coatings applied at increasing reaction temperature (25-95 °C) are consistent with d increasing from 2.4 to 11 nm (Table 2). It appears that an increase in packing density may also increase polymer layer thickness. An increase in coating thickness with polymer grafting density is in keeping with recent theory.19,41-43 Lim and Herron reported modeling experiments in which the thickness of coatings of similar MW PEG varied with polymer packing density and PEG-solvent interactions.42 Malmsten and Van Alstine used ellipsometry to study adsorption of linear PEG-acyl surfactants at methylsilanemodified quartz. The PEG surfactants rapidly formed saturated monolayers 10-15 nm thick at packing densities approaching 0.1 chains‚nm-2. Repulsive interchain interactions appeared to determine packing density and cause elongation of PEG molecules normal to the surface.44 The effects of reaction time and pH on E-PEG grafting density were also studied. Varying reaction time from 6 to 24 h minimally reduced electroosmosis (not shown), while varying reaction pH from 7.3 to 10.3 gave similar results (Figure 7). Given that enhanced reaction time and reactivity (at more basic pH) do not (41) de Gennes, P. J. Macromolecules 1980, 13, 1069-1075. (42) Lim, K.; Herron, J. N. In Poly(ethylene glycol) Chemistry: Biotechnical and Biomedical Applications; Harris, J. M., Ed.; Plenum: New York, 1992; pp 29-56. (43) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.; Cosgrove, T.; Vincent, B. Polymers at Interfaces; Chapman and Hall: London, 1993. (44) Malmsten, M.; Van Alstine, J. M. J. Colloid Interface Sci. 1996, 177, 502512.

3756 Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

significantly increase grafting density, the temperature effect noted above does not appear to be related to thermal acceleration of the reaction. Surface-localized PEGs should hinder grafting due to repulsive interchain interactions. Ellipsometric studies suggest that at 25 °C, such interactions might become significant below 0.1 chains‚nm-2.44 PEG molecular conformation is known to be influenced by temperature. As reaction solution temperature approaches the PEG lower critical solution temperature (θ, approximately 105 °C under these conditions45), it is expected that the PEG radius of gyration and interchain repulsion are reduced, reactive functional groups become more accessible, and nonspecific adsorption is enhanced. Such conformational effects have been used to enhance surface grafting of ethylhydroxyethylcellulose22 and enzyme tethering via ethylene oxide-poly(propylene oxide) copolymer.46 To test if such effects might explain the observed relationship between grafting density and temperature, grafting was performed at 35 °C (approximate θ45) for 6 h, pH 6.9 in 0.4 M K2SO4. The electroosmosis profiles and modeling (Table 2) for such coatings mimicked those obtained at 60 °C in 0.05 M sodium phosphate solution (Table 2, plus Figures 5 and 6). The radius of gyration (Rg) of PEG 3400 in free solution is ∼3.5 nm.41,47,48 This is larger than the 2.4 nm d value found for the less dense covalent PEG coatings (Table 2) or independent thickness estimates for adsorbed PEG.49 While caution should be exercised in relating d values to polymer surface conformation,29,43 this result suggests interaction between surface silanol groups and PEG.50 A variety of studies suggest that PEG, poly(vinyl alcohol), and other neutral polymers may adsorb to such silanols and reduce the expression of surface potential without altering the surface density of such groups.5,10,30,49 Such a surface-PEG interaction is also in keeping with observed increases in silanol group pK.18,48-51 Exposing clean quartz capillaries to underivatized PEG of 25 000 Da resulted in silanol pK shifts similar to those noted for capillaries grafted with E-PEG (Table 2). In both cases, experimental results were best modeled by paired shifts in isolated and associated silanol pK from 3.6 to 4.1 and 6.9 to 8.0 (Table 2, Figures 5 and 6). A variety of other pK shifts (e.g., 3.6 and 8.0, or 4.1 and 6.9) gave less accurate modeling. The only exception was that the E-PEG coating obtained at 95 °C was slightly more accurately modeled using silanol pK values of 3.6 and 6.9. This may reflect reduced PEG-silanol interactions at high grafting density or merely the linear nature of the pH versus electroosmosis profile (Figure 6). One of a variety of possible PEGsilanol interactions is shown in Figure 4a. The modeling assumption of various silanol groups with distinct pK values in Tables 1 and 2 or in the results of other authors (see introduction) may, in fact, represent a broad range of silanol environment and behavior, as opposed to distinct molecular entities. (45) Elias, H. G. In Polymer Handbook, 3rd ed.; Brandrup, J., Immergut, E. H., Eds.; Wiley Interscience: New York, 1989; Chapter VII, pp 205-231. (46) Tiberg, F.; Brink, C.; Hellsten, M.; Holmberg, K. Colloid Polym. Sci. 1992, 270, 1188-1193. (47) Devanand, K.; Selser, J. C. Macromolecules 1991, 24, 5943-5947. (48) Gabble, A.; Horn, R. G. J. Colloid Interface Sci. 1993, 157, 375-383. (49) Van der Beek, G. P.; Cohen Stuart, M. A. J. Phys. Fr. 1988, 49, 14491454. (50) Char, K.; Gast, A. P.; Frank, C. W. Langmuir 1988, 4, 989-998. (51) Eremenko, B. V.; Platonov, B. EÄ .; Uskov, I. A.; Lyubohenko, I. N. Kolloid. Z. 1974, 36, 240-244.

CONCLUSIONS Site dissociation modeling27 of electroosmosis over a broad range of pH was shown to provide insight into various surface chemical aspects of native, APS-modified, PEG-adsorbed, and PEGgrafted quartz. The results further emphasize the analytical value of this approach18,52 which, given the use of molecular versus particle flow markers, should be readily applicable to much smaller bore capillaries. Under the conditions studied, native quartz gained surface charge over a broad range of pH. This could be interpreted on the basis of isolated and associated silanol groups of pK 3.6 and 6.9, respectively. Silanol densities varied with quartz history from 0.3 to 5.0 groups‚nm-2. The conditions described yield an APS coating of ∼0.2 charge groups‚nm-2, which was stable for at least 6 h at 60 °C in 0.05 M salt, pH 10.3. The grafting of E-PEG to this surface was relatively independent of pH (7.310.3) and reaction time (6-24 h) but was significantly influenced by reaction temperature and salt composition. This suggests that PEG solution conformation influences PEG grafting density. Surface-localized PEG molecules appear to interact with quartz silanol groups and increase silanol group pK. The significant reduction in electroosmosis over a broad pH range from grafting E-PEG 3400 appears to be primarily due to creation of a polymerenriched layer that opposes fluid flow up to 10 nm from the (52) Marlow, B. J.; Rowell, R. L. Langmuir 1991, 7, 2970-2980.

surface. Previous research involving APS- and mercaptopropylsilane-based coatings5,18 suggests that these conclusions are not limited to APS-E-PEG coatings. In keeping with other studies, the thickness of terminally anchored PEG coatings appears to increase with grafting density. Manipulation of reaction solution temperature and salt composition allows control over PEG grafting density. Relatively dense (surface masking) PEG coatings are obtainable under conditions sufficiently benign for tethering proteins and other sensitive molecules. Several lines of research are indicated, including studies concerned with the stability, protein adsorption, and tethering properties of PEG coatings of different grafting densities. ACKNOWLEDGMENT The authors thank Norman Burns and Martin Malmsten for discussion. This work was supported by the National Science Foundation EPSCoR Program (J.M.H. and J.V.A.) and the National Aeronautics and Space Administration (J.V.A.). J.V.A. acknowledges Wenner-Gren Foundation support at the Institute for Surface Chemistry, Stockholm. Received for review February 1, 1996. Accepted June 17, 1996.X AC960114M X

Abstract published in Advance ACS Abstracts, August 1, 1996.

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

3757